On Rate Fairness Maximization of Vehicular Networks: A Deep Reinforcement Learning Approach

Abstract

This paper investigates the rate fairness maximization (FM) in a vehicular network consisting of multiple vehicle-to-vehicle(V2V) pairs and vehicle-to-infrastructure (V2I) pairs. To this end, we formulate the FM as an optimal problem subject to the constraints of the quality of service (QoS) requirements, spectrum and power resources. It is usually challenging to solve this nonlinear and nonconvex optimization problem. To tackle with this challenge, we further model the spectrum sharing between V2V and V2I links, and the transmit powers of V2V and V2I users as a Markov decision process. Then, a deep reinforcement learning-based algorithm is proposed to maximize the rate fairness while meeting the constraints of the QoS requirements by jointly optimizing the allocations of spectrum and power resources. Finally, simulation results are presented to illustrate our findings.